Application of a neural network to granulation scale-up

This paper describes a practical method for granulation scale-up by means of a neural network. Wet granulation was conducted using an agitation fluidized bed, and the scale-up characteristics were investigated using a neural network with back-propagation learning. Granule properties obtained by production-scale granulation under various operating conditions were predicted. Extremely good correlation was obtained between the predicted data and the experimental data of agitation fluidized bed granulation. It was found that granulation scale-up could be conducted with high accuracy by a neural network without constructing a mathematical model with a complicated non-linear relationship using a vast amount of experimental scale-up data.